Report of the Round Table Session
Miksis-Olds, J.L.1*, Heaney, K.D.2*, Martin, B.3, Hawkins, A.4, Širović, A.5, Heise, K.6, Kaplan, M.7, and Mennitt, D.J.8
1 Applied Research Laboratory, The Pennsylvania State University, USA
2 OASIS Inc., USA
3 JASCO Applied Sciences, Canada
4 Loughine Limited, UK
5 Scripps Institution of Oceanography, UCSD, USA
6 Vancouver Aquarium and WWF Canada, Canada
7 Woods Hole Oceanographic Institution, USA
8 Colorado State University, National Park Service, USA
This report can be referenced as:
Miksis-Olds, J.L., Heaney, K.D., Martin, B., Hawkins, A., Širović, A., Heise, K., Kaplan, M., and Mennitt, D.J. (2015). Report of the Soundscapes Session, Oceanoise2015, Vilanova i la Geltrú, Barcelona, Spain, 10-15 May. (Editors Michel André & Peter Sigray). Retrieved from http://oceanoise2015.com
A soundscape, or auditory landscape, is a combination of the traditional measured physical sound signal and the dynamically changing acoustic environment. It is composed of multiple sound producing sources, the perception of which depends upon the relative contribution of each source, direction, and propagation of the signals. The soundscape can be selectively decomposed and visualized to gain a greater understanding of the sources and environmental dynamics contributing to and shaping the temporal, spatial, and spectral patterns of the acoustic environment. Displaying the complexity of these three defining soundscape aspects in relation to geophony, biophony, and anthrophony in a coherent, meaningful way is a challenge. We also need to develop a better understanding for how marine life perceives, responds to, and is impacted by different features of the constantly changing soundscape.
The field of underwater acoustics has struggled to define and apply the concept of soundscape to ocean research since its first appearance in the peer-reviewed literature in 1997. Potter et al. (1997) highlighted the spatial aspect of the underwater soundscape through a series of power spectrum figures reflecting representative spectra for different source types in different regions of coastal Singapore waters. Evolution of the underwater soundscape concept over the past two decades is evidenced by its academic expansion from 2-D representations of the physical environment via times series and spectra to 4-D interactive soundscape maps incorporating user specified source types, numbers, and acoustic characteristics. Underwater soundscapes are currently predominantly utilized in association with three topic areas: 1) signal detection, 2) marine life use and response to sound, and 3) indicators of ecosystem status or health. Application of the underwater soundscape concept has now been used in aquatic research across ecosystems ranging from shallow coastal bays and estuaries to the deep ocean and from physics based detection applications to defining its value through an ecosystem services approach (NRC, 2008).
Regardless of the application, those investigating and analysing underwater soundscapes face the same challenge – What is the best way to define and visualize the dynamic underwater soundscape? Sound travels extremely efficiently underwater, so sources thousands of kilometres away can greatly contribute to regional and local soundscapes given the right propagation conditions in deep water (Munk, 1974). This creates confusion and complexity in defining the spatial scope of an underwater soundscape. Similarly, the acoustic environment is constantly changing, so defining temporal scales of variability in connection with underwater soundscapes shifts in accordance with the scale of the question being investigated. Working with soundscape features in the spectral realm is also complicated, as source characteristics change over time (on a daily, weekly, seasonal, and annual basis). The non-linear, unpredictable, often unknown nature of biological sources greatly contributes to uncertainty in spectral analyses.
This session of the workshop highlighted advances in the temporal, spatial, and spectral approaches to underwater soundscape analyses. Technology developments supporting longer term, higher bandwidth recordings and faster, more efficient analyses have enabled parallel advances in the use and application of underwater soundscapes beyond that of qualitative descriptions. Temporal patterns of sound sources and types are being used to both characterize and compare soundscapes (Širović & Oleson, 2015; Kaplan & Mooney, 2015; Hawkins, 2015). Martin and Barclay (2015) are adding to the traditional method of detecting and classifying contributing sources of the underwater soundscape by detector algorithms with an innovative method utilizing temporal rhythms of source types on daily, weekly, and tidal scales.
In the spatial realm, simple soundscape comparisons of level or spectrum have evolved into multi-dimensional and directional information flows. In regions without high resolution anthropogenic, biological or bottom composition data related to sound contribution and propagation, Mennitt et al. (2015) described a geospatial modelling approach to create sound maps in the Stellwagen Bank National Marine Sanctuary, USA. Heaney (2015) illustrated how a deep water, vertical line array can be used to gain spatial information related to source directionality and what that could potentially mean to listening animals. Examination of the soundscape at multiple sites along the BC, Canada coast has also afforded the tracking of killer whale pods and humpback whales across space and time (Heise, 2015).
A majority of the presentations combined temporal and/or spatial investigations with spectral analyses to gain additional knowledge about the soundscape. Looking at the shifts in spectra over time at one location off Hawaii, USA, Širović & Oleson (2015) identified a frequency and unit shift in humpback song. Similarly, through the use of frequency correlation matrices Miksis-Olds & Nichols (2015) identified a shift in fin whale song structure in the Equatorial Pacific Ocean at Wake Island. Analyses utilizing the ratio of sound levels across the soundscape spectrum are now providing information about the biodiversity, condition of habitats, and status of animal populations that was not tractable through passive acoustic recordings just a decade ago (Širović & Oleson, 2015; Kaplan & Mooney, 2015; Hawkins, 2015).
Questions arising from the discussion panel covered a multitude of topics including sound measurement, processing, perception, and impact. Questions are grouped by topic below and reflect the current conflicts and challenges in the field of underwater soundscapes:
How do we address the scale over which a soundscape and its variability is representative? When taking measurements what sort of spatial scales are implied related to spatial correlation. How many recorders are needed to adequately summarize a soundscape? Observations of 4 receivers on a healthy reef (Kaplan & Mooney, 2015) showed differences among the sites were small. There was an observation in deep water (Širović & Oleson, 2015) where two hydrophones separated by a short distance off of an island had significantly different soundscapes. The problem can be initially addressed by separating out frequency bands, and regions where specific sources (ships, wind, snapping shrimp) dominate. There is a need to use modelling to predict the impact of masking, as well as to build an expectation of the spatial correlation based upon environment, frequency and source types.
What kind of averages should be taken when processing the data and how should we handle transients when averaging or when evaluating masking? Consensus was that the mean is a poor indicator of the soundscape and that the median or various percentiles provide a better understanding of the processes associated with the acoustic environment. Removing transients can exclude important sources, and trend analysis requires including transients. Studies to determine the effect of noise on masking of signals need to be performed with the appropriate time-bandwidth used for the soundscape evaluation. Transients can be important to the behavioural response of animals. It is best to quantify the occurrence rate of transients, their spectral characteristics, as well as the percentiles of noise to fully interpret the overall soundscape.
What is the soundscape from the perspective of the animal and how should we as a field adjust the analysis accordingly? We can learn from our terrestrial counterparts – humans and small mammals. It is recognized that animals use cues from the soundscape to aid in navigating through their environment. We do know fish are very good at navigating, often over great distances back to natal spawning grounds. It seems that many marine mammals use hearing to find their way around. Navigation is either by pilotage or directional noise cues. We do not know enough about geophysical acoustic cues that could be used. Animals aggregate around open ocean seamounts – acoustics could be a possibility for localization of deep seamounts. Animals may be able to use sound, and can be impacted by sound outside of their hearing ranges. Marine animals are multimodal – sight, hearing, chemical, lateral line/particle motion. Incorporating perception into the concept of underwater soundscapes is a critical and valuable step towards understanding the impact of soundscapes from an animal perspective.
The community, represented by this panel, has worked hard in the characterization space of soundscapes. How do we move towards the impact space? It would be very helpful to evaluate the potential for masking impacts in a signal agnostic way. This would involve characterization of impacts for disturbances at long range and over extended time and could address the potential impact of expected changes in source level distributions.
Acoustic ecology is about relationships, can you extend this analysis to predator – prey relationships? The bioacoustics community has been looking at habitat use by animals through acoustics. It is a challenge when spatial localization is difficult due to long propagation distances and minimal receiver aperture. We are just starting to see researchers move from the measurement to the ecological links – acoustic complexity index, entropy etc., similar to the terrestrial system evaluating biological diversity and quality. The acoustics community is just starting to do this underwater. It is more difficult due to the long-range nature of propagation. Ecosystem boundaries are much larger for acoustics than for chemical or physical characteristics. The engineering is advancing so we are capable of recording space-time data, but a vast majority of recordings continue to be single element hydrophones. We need long term data AND directional data. There is also a need to pair the temperature and chemistry to the ecosystem description.
Evolution has led to extensive use of sound by all animals, even more so for animals living in the acoustically transparent ocean. Hearing evolved so that animals can sense their 3-D world. 400 million years of evolution did not select for communication but for animals to sense their environment. If we impact the soundscape, we are harming the ability of animals (invertebrates, fish, and marine mammals) to fully sense the world in which they live.
The discussion panel concluded with a short interchange of ideas on immediate needs of the field. Two needs were highlighted: 1) better ways to visualize and compare soundscapes, and 2) long-term, continuous, full-bandwidth recordings. New research shared during the workshop demonstrated the progress the field of underwater acoustics has made in defining, measuring, processing, and visualizing underwater soundscapes, and the resulting discussion identified areas where new information is needed. The concept of underwater soundscapes continues to evolve, and the next 5-10 years will produce some exciting new research in this area.